Collection constructor is very slow with large lists

[hsolbrig]

The collection constructor uses the append method, which iterates to the end of the list for every addition. This results in a n! performance. Propose replacing:

    for item in seq:
         self.append(item)

with:

    if seq:
        nxt = self.uri
        for item in seq:
            graph.add((nxt, RDF.first, item))
            nxt2 = BNode()
            graph.add((nxt, RDF.rest, nxt2))
            nxt = nxt2
        graph.add((nxt, RDF.first, RDF.nil))

in the constructor

[joernhees]

it's O(n^2), but i agree that it's unnecessary... if you make a pull request, i'll accept it.

[gromgull]

I went for a fix without the _end caching issues. Now there is a __iadd__ method that lets you do my_collection += [ Literal(1), Literal(2) ] and this will be efficient. Repeat calls to append remain inefficient, but this is safer!